e-commerce & quick commerce pm
In e-commerce, your product is not the app. Your product is the promise — that the right thing will arrive at the right time for the right price. Every screen, every algorithm, every warehouse decision either keeps that promise or breaks it.
Most PMs who join an e-commerce company think they are building a shopping app. They obsess over search UI, product detail page layouts, checkout flows. Those matter. But the PM who only sees the screen will always lose to the PM who sees the system behind it.
In Indian e-commerce, the system is: a seller in Surat uploads a kurta image shot on a bedsheet. A buyer in Patna searches “party wear kurti” on their phone over a 3G connection. A warehouse in Bhiwandi ships it via a logistics partner who subcontracts to a delivery person on a motorcycle. The buyer pays cash on delivery. The kurti does not match the photo. The buyer refuses the package.
Every step in that chain is a product problem. And the PM who treats catalog quality, logistics, and payments as “ops issues” will spend their career building pretty interfaces on top of a broken system.
The three kinds of Indian e-commerce PM work
Indian e-commerce is not one industry. It is three, and the PM work differs sharply.
Horizontal marketplaces — Flipkart, Amazon India, Meesho. You are managing two-sided network effects: sellers and buyers. Your job is to make it easy for sellers to list and profitable for them to stay, while making it easy for buyers to find and trust what they are buying. The core tension: seller count vs. catalog quality. More sellers means more selection but also more garbage listings.
Vertical commerce — Nykaa (beauty), Lenskart (eyewear), Pepperfry (furniture). You control the supply chain deeper. Catalog quality is higher because you curate. But you have a harder growth problem: you cannot win on selection breadth, so you must win on expertise, trust, and experience. The PM work is closer to brand-building than marketplace management.
Quick commerce — Zepto, Blinkit, Swiggy Instamart. Everything about your product revolves around one number: delivery time. 10 minutes changes the economics, the assortment, the warehouse design, the delivery partner’s experience. You are building a logistics product that happens to have a consumer app on top of it.
If you are choosing between these three, know what you are signing up for. Marketplace PMs spend 60% of their time on seller-side and catalog problems that users never see. Vertical PMs spend 60% of their time on merchandising and brand experience. Quick commerce PMs spend 60% of their time arguing with operations about what is physically possible.
Marketplace dynamics are PM problems
Two-sided marketplaces have a specific PM challenge: every decision you make for one side affects the other. Improve seller onboarding and you get more listings, but if those listings are low quality, buyer trust drops. Crack down on catalog quality and sellers leave for a competitor with lower standards.
Seller onboarding is a product, not a form
At Meesho, getting the next million sellers online is the growth engine. At Flipkart, keeping high-quality sellers from leaving for Amazon is the retention challenge. Both are PM problems, not sales problems.
The seller onboarding flow — account creation, tax verification, catalog upload, first listing, first order — is a product funnel with the same conversion logic as any consumer funnel. Where do sellers drop off? At GST verification, because the process is confusing. At catalog upload, because photographing products is hard for a small shopkeeper. At first order, because it takes too long and they lose confidence.
Every one of those is a product intervention, not a training manual.
The buy box is where PM meets economics
On Amazon India, when multiple sellers offer the same product, one seller “wins” the buy box — the default “Add to Cart” option. The algorithm considers price, delivery speed, seller rating, and inventory reliability.
The PM who owns the buy box algorithm is arguably the most powerful PM in the company. A tweak to the weighting can shift crores of revenue between sellers overnight. This is where PM work, pricing strategy, and marketplace health intersect. Get it wrong — weight price too heavily — and sellers race to the bottom, quality drops, and buyers start seeing counterfeits. Get it wrong in the other direction — weight seller rating too heavily — and new sellers can never compete, and selection stagnates.
Search and discovery are not just engineering problems
On a marketplace with 150 million products, search is the product. A buyer who searches “red saree for wedding” and sees a page of irrelevant results will leave. A buyer who searches “iPhone 15 case” and sees a sponsored result they did not ask for will lose trust.
Search quality on Indian e-commerce has specific challenges. Vernacular queries (buyers searching in Hindi or Tamil or Bangla). Spelling variations that are not typos — they are how people in a specific region spell the word. Attribute-sparse listings where the seller wrote “good quality kurta” instead of specifying fabric, colour, and size.
The PM owning search is working on an NLP problem, a catalog quality problem, and a user trust problem simultaneously.
Catalog quality: the product problem nobody wants to own
Here is the uncomfortable truth about Indian e-commerce. The catalog is the product. Not the app, not the recommendation engine, not the checkout flow. The catalog.
When a buyer opens Flipkart and sees a product image that was clearly photographed on a kitchen counter with bad lighting, and the description says “Best quality product material very good buy now,” that is not a seller education problem. That is a product failure.
At scale, you cannot manually fix 150 million listings. You need:
Automated quality scoring — image quality detection (resolution, background, watermarks), description completeness checks, attribute extraction from unstructured text. The PM building this is working on an ML product.
Seller tooling — give sellers a phone-based photo studio in the app. Guide them through background removal, multi-angle shots, and attribute tagging at the point of upload. Meesho invested heavily here because their sellers are small shopkeepers, not professional retailers.
Duplicate detection — the same product listed 47 times by different sellers with different titles, different images, and different prices. Catalog deduplication is one of the hardest unsolved problems in Indian e-commerce. The PM who cracks it will save their company hundreds of crores in wasted ad spend and buyer confusion.
A common mistake: treating catalog cleanup as a one-time ops project. It is not. New low-quality listings arrive daily. Catalog quality is an ongoing product system with automated enforcement, seller feedback loops, and continuous ML model improvement.
Weekly business review at a quick commerce company. The new dark store in Indiranagar, Bangalore is underperforming.
Head of Expansion: “We should open two more dark stores in South Bangalore. Indiranagar is capacity-constrained during peak hours — we are losing orders because we cannot fulfil fast enough.”
Quick Commerce PM: “Before we open new stores — have we looked at what Indiranagar is actually stocking? I pulled the data yesterday. 40% of the shelf space is allocated to SKUs that sell less than 5 units a week. Meanwhile, we are stocking out on eggs, milk, and bread by 11 AM every day.”
Head of Expansion: “Assortment is an ops problem. I am talking about coverage.”
Quick Commerce PM: “It is the same problem. If we fix assortment in the existing store, we free up capacity for 30% more orders without any capex. Opening a new store costs 40 lakhs. Fixing the planogram costs an analyst's time for two weeks.”
Head of Operations: “She has a point. We opened Koramangala with the same generic assortment and it took three months to stabilise. Let me pull the demand prediction data for Indiranagar.”
The expansion decision was paused. Within two weeks, the assortment fix increased Indiranagar's daily orders by 22% without a single new dark store.
The instinct is always to expand coverage. The disciplined PM asks whether the existing footprint is performing at potential first.
Quick commerce: when 10 minutes is the entire product
Quick commerce changed Indian e-commerce in a way most people underestimate. When Zepto, Blinkit, and Swiggy Instamart promised delivery in 10 minutes, they did not just speed up grocery shopping. They eliminated the need to plan. You do not buy groceries for the week when you can get anything in 10 minutes. You buy what you need right now.
This has massive product implications.
The dark store is the product
A dark store is a small warehouse — 2,000 to 3,000 square feet — optimised for picking speed, not browsing. The PM working on dark store operations is solving:
Assortment optimisation — which 2,000 SKUs (out of 50,000+ possible) should this specific dark store carry? The answer differs by neighbourhood. A dark store in Bandra needs different products than one in Noida. Demand prediction, hyperlocal purchasing patterns, and seasonal adjustments are all PM decisions.
Planogram design — where each product sits on the shelf affects picker speed. If eggs are at the back and bread is at the front but 80% of orders contain both, the picker walks the full length of the store for every order. The PM who reduces average pick time by 30 seconds across a million orders a day has saved the company more money than anyone working on the app.
Demand prediction — if you stock too much, perishables expire and you eat the loss. Stock too little, and you show “out of stock” on the app, which trains buyers to open a competitor instead. The prediction window is 4-6 hours because dark stores get replenished multiple times a day.
Delivery partner experience is a product surface
The delivery partner — usually a person on a motorcycle or bicycle — is part of the product. If their app is confusing, they take longer. If the route optimisation is wrong, they take longer. If the batching algorithm assigns them three orders in opposite directions, they take longer. Every minute they waste is a minute the customer waits.
Quick commerce PMs who ignore the delivery partner experience are optimising the wrong end of the funnel. The partner’s app, the order batching logic, the navigation accuracy in Indian lane-and-gully addresses — these are product surfaces that directly determine whether you hit 10 minutes or not.
The unit economics problem nobody wants to talk about
Here is the tension that defines every quick commerce company in 2024-25. The 10-minute promise costs money: small dark stores have higher rent per square foot, frequent replenishment runs cost more than weekly bulk deliveries, delivery partners are paid per order and need to be close to the store.
At current basket sizes — average order value of Rs 350-450 — most quick commerce orders lose money. The path to profitability runs through: increasing AOV (pushing higher-margin categories like personal care and snacks alongside groceries), increasing order density per dark store (so fixed costs spread over more orders), and reducing delivery cost per order (batching, route optimisation, hub-and-spoke models).
Every one of those is a product problem. The PM who can increase AOV by 15% through better recommendation and bundling — without annoying the buyer who just wants milk — has solved a problem worth hundreds of crores.
Logistics as a product constraint
In most tech companies, logistics is something that happens after the product decision. In Indian e-commerce, logistics constrains the product decision.
Pincode serviceability shapes the entire experience
Before a buyer sees a product on the app, the system has already checked: can we deliver this product to this pincode? The answer depends on the seller’s location, the logistics partner’s coverage, the product’s size and weight, and whether the destination is COD-eligible.
A buyer in Ranchi searching for a refrigerator sees a different catalog than a buyer in Mumbai searching for the same thing. Not because the products do not exist — but because the logistics chain cannot deliver a refrigerator to certain pincodes at an acceptable cost and timeframe.
The PM working on serviceability is not solving a logistics problem. They are solving a selection problem. Every unserviceable pincode is a customer you cannot convert.
Cash on delivery: the Indian e-commerce tax
COD accounts for 40%+ of Indian e-commerce transactions. In Tier-2 and Tier-3 cities, it is 60%+. This is not a payments preference — it is a trust signal. Buyers in smaller cities do not trust that the product will match the listing. COD is their insurance policy: see the product, then pay.
COD has real costs. Higher return rates (buyers treat COD as a free trial). Cash handling and reconciliation at the last mile. Fraud — fake orders placed to harass someone at an address, or repeated order-and-refuse cycles.
The PM who reduces COD share from 55% to 40% in a region — through UPI incentives, prepaid discounts, trust signals on the product page, or a COD-to-prepaid graduation model — has improved margins more than any checkout flow optimisation ever will.
Returns are a product system, not a cost centre
Returns in Indian e-commerce run 15-25% depending on the category. Apparel is worse — 30%+ because sizing is inconsistent across brands and the product often does not match the listing image.
Most companies treat returns as a logistics cost to minimise. The better framing: returns are a signal. Every return tells you something — the listing was inaccurate, the size chart was wrong, the delivery was late, or the buyer never intended to keep it (wardrobing, COD abuse).
The PM who builds a returns intelligence system — categorising return reasons, feeding them back to catalog quality, seller scoring, and size recommendation — turns a cost centre into a product improvement engine.
Vernacular commerce: Meesho’s lesson
Meesho did something that Flipkart and Amazon India struggled with for years. They built for the buyer who does not search in English, does not browse categories, and does not trust online shopping.
Their insight: in Tier-2 and Tier-3 India, commerce is social. A woman in Lucknow sees a kurta on WhatsApp, shared by someone she trusts, and wants to buy it. She does not want to open an app, search for it, compare prices, and add to cart. She wants to tap the image and buy.
This is a fundamentally different product model:
Social sharing as discovery. Instead of search and browse, products spread through WhatsApp and social networks. The PM work is making sharing easy, making the shared link convert, and making the reseller model profitable for the person sharing.
Regional language as default. Not a “translate” button in the corner — the entire experience, including seller communication, order tracking, and customer support, in the buyer’s language. This is an infrastructure decision, not a UI decision.
Price-first assortment. Meesho’s buyer is not looking for brands. They are looking for the cheapest saree that looks decent. The product ranking algorithm is fundamentally different from Amazon’s — price and value dominate over brand and reviews.
If you are building for India’s next 500 million internet users, study Meesho’s product decisions carefully. They solved problems that most e-commerce PMs do not even recognise as problems because they are building for a buyer who looks like themselves.
Pick any Indian e-commerce app you use — Flipkart, Amazon, Meesho, Zepto, Blinkit, Nykaa, whatever you order from most.
Place a real order (or recall your most recent one). Map the end-to-end flow from the buyer’s perspective:
- Discovery: How did you find the product? Search? Recommendation? Category browse? External link?
- Evaluation: What information did the product page give you? What was missing? Did you check reviews? Compare with other sellers?
- Purchase: How many taps from “Add to Cart” to order confirmation? Was the payment flow smooth? Were there upsells or nudges?
- Fulfilment: How accurate was the delivery estimate? Did you get tracking updates? Were they useful or just noise?
- Post-delivery: Was the product what you expected? If not, what went wrong in the listing? Did you consider returning it?
Now identify the three biggest friction points in that flow. For each one, answer:
- Is this a product problem (the app could be better), a catalog problem (the listing was wrong), or a logistics problem (the delivery was unreliable)?
- Who owns this problem in the company? If the answer is “nobody” or “it falls between teams,” you have found the most important problem to solve.
You are the quick commerce PM at Zepto. You have been piloting a dark store in Nagpur for 4 months. Average order value is ₹320, delivery time is 11 minutes, and you are processing 180 orders per day. The unit economics show a net loss of ₹28 per order after accounting for dark store rent, picker wages, and delivery partner cost. The Nagpur market has strong demand signals — 40% of orders are repeat customers — but the financials do not close at current basket sizes. Mumbai and Pune dark stores break even at 280 orders per day at ₹380 AOV. Nagpur is structurally different: smaller basket sizes, lower purchasing power per order.
The call: Do you keep the Nagpur dark store open, or close it and focus on markets where the unit economics work?
You are the quick commerce PM at Zepto. You have been piloting a dark store in Nagpur for 4 months. Average order value is ₹320, delivery time is 11 minutes, and you are processing 180 orders per day. The unit economics show a net loss of ₹28 per order after accounting for dark store rent, picker wages, and delivery partner cost. The Nagpur market has strong demand signals — 40% of orders are repeat customers — but the financials do not close at current basket sizes. Mumbai and Pune dark stores break even at 280 orders per day at ₹380 AOV. Nagpur is structurally different: smaller basket sizes, lower purchasing power per order.
The call: Do you keep the Nagpur dark store open, or close it and focus on markets where the unit economics work?
What e-commerce PM careers actually look like
E-commerce PM roles in India cluster into a few tracks, and they require different skills.
Buyer experience PM — search, discovery, product detail pages, checkout. This is the track most PMs think they are signing up for. It is important work, but it is also the most competitive and the least differentiated. Your search ranking skills at Flipkart are the same skills needed at Amazon or Meesho.
Seller experience PM — onboarding, catalog tools, seller analytics, pricing tools, seller growth programs. Less glamorous, often higher impact. The seller side of a marketplace is chronically underinvested because leadership attention goes to the buyer side. This means more room to drive results.
Catalog and content PM — quality scoring, deduplication, image and description standards, automated enrichment. This is increasingly an ML-heavy role. If you like working at the intersection of data quality and product, this is where the interesting problems are.
Logistics and fulfilment PM — serviceability, delivery experience, returns, warehouse operations. You will work more with operations than engineering. The problems are physical-world constrained. If you enjoy systems thinking with real-world constraints, this is the track.
Payments and risk PM — checkout optimisation, COD reduction, fraud detection, payment success rates. Highly analytical. A 0.5% improvement in payment success rate on millions of daily transactions is worth crores.
Quick commerce PM — dark store operations, demand prediction, delivery optimisation. A mix of all of the above but compressed into a 10-minute window. The feedback loops are the fastest in any PM role — you can see the impact of a change within hours, not weeks.
Test yourself
You are a PM at a quick commerce company. Average delivery time is 12 minutes — solid, but the CEO has seen a competitor claim 8-minute delivery in select areas. She wants 8 minutes across all dark stores within 3 months. Your analysis shows that hitting 8 minutes with the current model would require either halving the delivery radius (cutting coverage by 40%) or doubling delivery partners (increasing costs by Rs 18 per order on a Rs 22 margin). The cost model breaks.
The CEO is in the room. The operations head looks uncomfortable. The CFO is staring at you. What do you do?
your path
Career-stage considerations
0-2 years: Start on buyer experience — search, discovery, product pages, checkout. This is where you get the highest volume of learning because the feedback loops are fast, the metrics are clear, and you develop intuition for how millions of users interact with a product. Do not chase the “sexy” logistics or ML roles yet.
3-5 years: Pick a track — catalog, logistics, seller experience, payments, or quick commerce — and go deep. The e-commerce PM who has done “a bit of everything” plateaus at mid-level. The one who deeply understands seller economics, or demand prediction, or payment success rate optimisation becomes the expert that companies recruit for senior roles.
5+ years: The e-commerce PM leader connects the supply chain to the user experience. At this level, you are not optimising one part of the system — you are seeing how catalog quality affects search results, how logistics constraints shape the product page, and how payment success rates determine whether a market is viable. This cross-system thinking is what makes Directors and VPs in e-commerce.
Where to go next
- Apply growth frameworks to e-commerce funnels: Growth Product Management
- Optimise the first-time buyer experience: Activation Optimization
- Build repeat purchase loops: Retention Loops
- Think through pricing and monetisation: Pricing Strategy